[FEATURE] Live Prompt Adjustment
Opened this issue · 1 comments
Problem Statement
Production agents need real-time optimization based on user feedback and A/B testing without service interruption. Every prompt improvement requires full redeployment, causing downtime and preventing rapid iteration on agent performance
Proposed Solution
Runtime modification of the agent's system prompt and reasoning instructions with thread-safe context preservation - updates take effect on next LLM invocation while maintaining conversation memory state.
Use Case
Runtime modification of the agent's system prompt and reasoning instructions with thread-safe context preservation - updates take effect on next LLM invocation while maintaining conversation memory state.
Alternatives Solutions
No response
Additional Context
No response
Hi! This can currently be accomplished via hooks. Checkout the BeforeModelCallEvent